Project Summary/Abstract Value-based insurance design (VBID) is widely viewed as an important tool to improve the health of consumers and the efficiency of health care spending in the U.S. VBID has found its widest applicability with respect to prescription drugs, and the potential of VBID may be greatest for elderly persons with chronic illnesses such as diabetes and cardiovascular disease that are effectively treated with prescription drugs. The essential feature of a VBID is the fact that prescription drugs may be substitutes or complements with other healthcare services. Therefore, changes in the use of prescription drugs, for example, because of changes in cost sharing, may affect health and alter use of other healthcare services. While conceptually appealing, the evidence to support VBID is lacking. Specifically, an effective VBID requires knowledge of the following: the effect of cost sharing on prescription drugs on the use of prescription drugs; the effect of cost sharing on prescription drugs on use of other healthcare services; and the effect of cost sharing on prescription drugs on health. While some evidence related to each of these links in the causal chain exists, it is not sufficiently detailed to design an effective ?value-based? prescription drug benefit plan. Our proposed research will provide evidence that will fill in much of the existing knowledge gaps and be instrumental in advancing the scientific base of evidence to inform future VBID, for example, as envisioned in the ?Model Test? program initiated by CMS. The proposed research will exploit the large and widespread changes over time in consumer cost-sharing within Medicare Part D plans to estimate the causal association between cost sharing and use of prescription drugs among the elderly, particularly among those with chronic illnesses such as diabetes and cardiovascular disease. We will then exploit the changes in consumer cost sharing to estimate the causal association between cost sharing and use of health care services other than prescription drugs, and cost sharing and health. These aims directly address the specific hypotheses described in PA-17-088: ?Differences in health outcomes between alternative treatment regimens or health care management strategies for older patients with specific common conditions in old age, or with specific combinations of two or more chronic conditions.? The research also focuses on the target population of PA-17-088: ?Groups identified by disease or non-disease status for comparison analyses (e.g., hip fracture patients) either retrospectively for risk factors (e.g., case-control studies) or prospectively for health-related outcomes. Groups defined by administrative databases to explore specific hypotheses regarding aging changes across the lifespan or diagnosis and management of medical conditions common among the elderly (e.g., CMS data, managed and/or accountable care organization data, health insurance databases, electronic health records, etc.).?